Evaluating The Feasibility Of Using Downwind Methods To Quantify Point Source Oil And Gas Emissions Using Continuous Monitoring Fence-line Sensors

Mercy Mbua², Stuart N. Riddick3,  Elijah Kiplimo²,  Kira Shonkwiler1, Anna Hodshire², and Daniel Zimmerle¹

¹CSU Energy Institute, Colorado State University, CO, 80523, Fort Collins, USA
²Department of Systems Engineering, Colorado State University, CO, 80523, Fort Collins, USA
3Department of Science, Engineering and Aviation, University of the Highlands and Islands Perth, Crieff Road, Perth, PH1 2NX, UK

Download Poster PDF

Background

Continuous monitoring solutions in oil & gas sites: detect methane CH4 & quantify
Captures temporal variation compared to survey methods

An aerial view of an oil and gas site with stars marking specific locations. Credit: Colorado Independent
Methane Continuous Monitoring Solution: Sensor, Deployment, Analytics

Quantification using point sensors is still uncertain:

  • Ilonze et al. (2024 reported emissions being misestimated by a factor of 0.2 to 42 for releases between 0.1 and 1 kg CH₄ h⁻¹, and by a factor of 0.08 to 18 for emissions above 1 kg CH₄ h⁻¹.

Objective: Evaluate the overall quantification accuracy of:

  • Eddy covariance
  • Gaussian Plume Inverse model (GPIM)
  • Backward Lagrangian stochastic model (bLs)

In quantifying single-release single-point and multi-release single-point emissions that simulate oil and gas emissions


Methodology

Experiments conducted at CSU METEC facility between February 8, and March 20, 2024.

A map of the METEC site showing the location of specific equipment and a closer up photo of the Sonic anemometer, 3 m analyzer inlet, and MGGA analyzer.

Quantification:

  • Eddy Covariance (EC): EddyPro® version 7 for flux & covariance footprints were calculated using the Kljun et al. (2015) and the Kormann and Meixner (KM) (2001) footprint models
  • bLs – WindTrax 2.0

GPIM equation:

GPIM equation where Q = mean sub E sub n sub h (x,y,x) * 2𝞹σ sub y σ sub z U all over exp (- y squared)/ 2σ sub y squared) * (exp ( - (z - H) squared/ 2σ sub z squared) + exp ( - (z + H) squared/ 2σ sub z squared))

Results

  • Our attempt to use eddy covariance (EC) failed because of problems with data collection and instrument performance. This produced invalid results, including unrealistically low emissions, large negative fluxes, and signal patterns (cospectra and ogives) that did not match expected shapes.
  • Consequently, the EC results could not be compared with the GPIM or the bLS models.
  • The bLs model demonstrated the highest accuracy for single-release single-point emissions, though it exhibited greater uncertainty than GPIM under multi-release conditions.
  • GPIM mostly overestimated emissions
Reference caption
Top plot: Estimated emission vs actual emission (kg h-1) for each model. The black dotted line is a 1:1 line based on actual emissions i.e. points below the line are underestimated and above are overestimated emissions. The gray region represents ±30% of the actual emission. Adj. R2 is the adjusted R2. The sample size is n. Bottom plot: Estimated emission vs actual emission categorized by Emission Point.

Conclusions and Next Steps

  • Oil and gas point sources could either be single emissions or multiple emissions occurring concurrently.
  • Single emissions are easier to quantify as they meet the models’ assumptions
  • For multiple sources, interference from neighboring emissions introduce ambiguity in model-observation alignment, particularly under complex wind conditions.
  • Closed-path eddy covariance was generally unreliable in this study due to data-collection and instrumentation issues associated with using a non-standard EC system.
Eddy Covariance diagram
  • For both GPIM and bLs, 15-minute averaging with a narrow wind-sector (5°) yielded the best performance.
Gaussian plume inverse model and Backward Lagrangian stochastic model with crosswind distance (m) downwind distance (m) point source and measurement point.

Acknowledgments and Contact Information

This work is funded by the Office of Fossil Energy and Carbon Management within the Department of Energy as part of the Site-Aerial-Basin Emissions Reconciliation (SABER) Project #DE-FE0032288. The authors thank Ryan Brouwer, Daniel Fleischmann, Ryan Buenger, and Wendy Hartzell for their assistance.

Poster author:

Mercy Mbua | Doctoral Student | Systems Engineering, Colorado State University [email protected] | LinkedIn


References

Mbua, M., Riddick, S. N., Kiplimo, E., Shonkwiler, K.B., Hodshire, A. & Zimmerle, D. (2025). Evaluating the feasibility of using downwind methods to quantify point source oil and gas emissions using continuous monitoring fence-line sensors. https://doi.org/10.5194/egusphere-2024-3161

This work is currently (October 2025) in the publication stages in the Atmospheric Measurement Techniques Journal.

Equipment at the METEC Site